module 'tensorflow._api.v1.compat.v2' has no attribute '__internal__' google colab error

Question

I am running a tensorflow model on google colab. Today, I got this error:

 Using TensorFlow backend.
    Traceback (most recent call last):
      File "train.py", line 6, in <module>
        from yolo import create_yolov3_model, dummy_loss
      File "/content/drive/MyDrive/yolo/yolo_plz_work/yolo.py", line 1, in <module>
        from keras.layers import Conv2D, Input, BatchNormalization, LeakyReLU, ZeroPadding2D, UpSampling2D, Lambda
      File "/usr/local/lib/python3.7/dist-packages/keras/__init__.py", line 3, in <module>
        from . import utils
      File "/usr/local/lib/python3.7/dist-packages/keras/utils/__init__.py", line 26, in <module>
        from .vis_utils import model_to_dot
      File "/usr/local/lib/python3.7/dist-packages/keras/utils/vis_utils.py", line 7, in <module>
        from ..models import Model
      File "/usr/local/lib/python3.7/dist-packages/keras/models.py", line 10, in <module>
        from .engine.input_layer import Input
      File "/usr/local/lib/python3.7/dist-packages/keras/engine/__init__.py", line 3, in <module>
        from .input_layer import Input
      File "/usr/local/lib/python3.7/dist-packages/keras/engine/input_layer.py", line 7, in <module>
        from .base_layer import Layer
      File "/usr/local/lib/python3.7/dist-packages/keras/engine/base_layer.py", line 12, in <module>
        from .. import initializers
      File "/usr/local/lib/python3.7/dist-packages/keras/initializers/__init__.py", line 124, in <module>
        populate_deserializable_objects()
      File "/usr/local/lib/python3.7/dist-packages/keras/initializers/__init__.py", line 49, in populate_deserializable_objects
        LOCAL.GENERATED_WITH_V2 = tf.__internal__.tf2.enabled()
      File "/usr/local/lib/python3.7/dist-packages/tensorflow_core/python/util/module_wrapper.py", line 193, in __getattr__
        attr = getattr(self._tfmw_wrapped_module, name)
    AttributeError: module 'tensorflow._api.v1.compat.v2' has no attribute '__internal__'

Previously, things had been running smoothly, so I'm not sure why this happened. I am using Python 3.7.10, and these are the packages I am supposed to use:

absl-py==0.9.0
astor==0.8.1
gast==0.2.2
google-pasta==0.1.8
grpcio==1.26.0
h5py==2.10.0
Keras==2.3.1
Keras-Applications==1.0.8
Keras-Preprocessing==1.1.0
Markdown==3.1.1
numpy==1.18.1
opencv-contrib-python==4.1.2.30
opt-einsum==3.1.0
protobuf==3.11.2
PyYAML==5.3
scipy==1.4.1
six==1.14.0
tensorboard==1.15.0
tensorflow==1.15.0
tensorflow-estimator==1.15.1
termcolor==1.1.0
tqdm==4.41.1
Werkzeug==0.16.0
wrapt==1.11.2

Perhaps colab recently upgraded some libraries? I am sure that I followed the same installation steps as I usually do.

EDIT: I think there may be an issue in the keras version. Here are the first few lines of the file I am running:

from keras.layers import Conv2D, Input, BatchNormalization, LeakyReLU, ZeroPadding2D, UpSampling2D, Lambda
from keras.layers.merge import add, concatenate
from keras.models import Model
from keras.engine.topology import Layer
import tensorflow as tf

If I remove all of the lines starting with "from keras", I don't get the error. However, I never touched these lines before, so I don't know why they would suddenly cause an error now. Also, it is not the python version causing this error, because colab changed it to 3.7.10 in April and I had no problem.

Solution 1

Try these parameters, it works with me

!pip3 uninstall keras-nightly
!pip3 uninstall -y tensorflow
!pip3 install keras==2.1.6
!pip3 install tensorflow==1.15.0
!pip3 install h5py==2.10.0

Solution 2

One morning I woke up and my code was giving the exact same error (in google colab).

I was running this code to downgrade my tensorflow

!pip install tensorflow==1.13.0

I replaced it with:

%tensorflow_version 1.x

Everything else remains the same. It worked!

Comments

[ansev]

I have the same problem

[Tahseen Adit]

I am facing the same problem in the middle of my thesis. It was working fine few days ago and then suddenly getting this error now. It is driving me crazy.

[Alexander Higgins]

Keras is clearly throwing the issue. I would open up a github issue on Tensoflow and Keras: ` File "/usr/local/lib/python3.7/dist-packages/keras/initializers/_‌​_init__.py", line 49, in populate_deserializable_objects LOCAL.GENERATED_WITH_V2 = tf.__internal__.tf2.enabled()`

[TFer2]

This issue is due to recent colab was upgraded to TF 2.5.0, forcing an upgrade to keras-nightly. For more information you can refer here. Thanks!

[blondelg]

Welcome to SO, please provide more details

[ddd]

Wow. This actually worked. Do you know why this happens?

[keramat]

ModuleNotFoundError: No module named 'keras.models'

[duhaime]

Colab loads tensorflow 2 by default. You have to use that %tensorflow_version 1.x declaration to use 1.x